Forum Space for engineering questions

[david]

Load dump is certainly the killer spec to meet due to its duration - typically several hundred mS making the use of transient protection devices impossible.  ISO 7637-2 lists 4 levels of test severity that must be quoted when listing compliance and these range from level one at 70V to level 4 at 200V.  If you have to meet this standard the best approach is often a high voltage FET and drive circuit to isolate the supply from the load while a transient protection device copes for the first few mS before the isolation kicks in.  The good news is that the test only has to be done once unlike some of the tests that are repeated thousands of times.

David

[John Lawton]

It is quite possible to protect against a short duration 200V transient simply. Split the feed resistor to the input diodes into two resistors and use a chunky (1W or more) zener diode to clamp the voltage to (say) 30V in the event of a spike. The zener diode and first input resistor need to be sized power wise to cope with the spike energy, the second resistor is sized to limit the opto diode current when 30V is present across the zener diode.

Something like:

[John Lawton]

As an afterthought, add a capacitor across the zener for both spike protection and contact bounce protection.

[See_Mos]

I think John's values are a bit high. at 24v input LED current is about 1mA or 0.5mA at 12v input.

On the TIG welding robots we built the cables ran right next to the welding cables which were between 5M and 25M long. With 24v for the switches etc. I just used a single resistor and capacitor in front of the LED and never had any problems.

[John Lawton]

Yes, they need to be quite high to prevent excessive current in the zener. The opto-detector load resistor would therefore need to be resized to match, bearing in mind the current transfer ration of the opto unit.

[david]

Sorry all - my previous post was about protecting the entire supply to this piece of electronics rather than just the logic inputs.  Apologies for any confusion caused.

David

[ken_k]

Hi all lots of good suggestions, I suggested an optocoupler with a bidirectional LEDs as it saved components on the board and negative going spikes exist, the alternative is an extra diode, with 0.6V forward drop they can handle a lot of current.
I have prepared an LTspice simulation that will work with components in the LTspice library.
change
opto_truck_load_dump_pulse.txt
to
opto_truck_load_dump_pulse.asc
and it should run.

I was intending to do more, I'm not feeling that well so will call it a night. I see a 4N25 LED is rated for an input current of 3A for 10us, there must be a derating that can be applied all the way down to its 60mA continuous current.

[John Lawton]

With a load dump spike of over 200V for over 350ms the 10us short term max current is not relevant as you say. I don't know whether a linear derating would apply, but I would suggest my circuit with a chunky zener diode would absorb more spike energy that a small LED ever could.

[JonW]

Without a cap the zener may well be too slow to protect against any high-speed transients/ESD, the rise times of these can be extremely fast, and thus you will need a low capacitance device.  A low voltage uni-directional TVS would be better than a Zener and designed specifically for this purpose, these will likely be the same price too.  I would be looking to use a high-speed TVS/ESD device depending on the transient.  The voltage across the LED will be limited to the VF of the LED anyway, so you should be able to use a USB-type device as these will start to breakdown around 7V which is << VF  of the opto and clamp any -ve excursions to the TVS VF.  If you use a dual Opto so the polarity can be reversed, then use a bidirectional TVS. NXP or littefuse do loads of TVS and ESD protection devices and they are cheap.

[John Lawton]

Good point about the cap to absorb any fast transients, the zener or TVS will then cope with any load dump pulses.

[ken_k]

With a load dump spike of over 200V for over 350ms the 10us short term max current is not relevant as you say. I don't know whether a linear derating would apply, but I would suggest my circuit with a chunky zener diode would absorb more spike energy that a small LED ever could.

Interestingly enough Fairchild rate their 4N25 Forward Current Peak at (3A, 300us, 2% duty cycle)
Toshiba also rate their 4N25 Forward Current Peak at (3 amps, 300us, 2% duty cycle).
To obtain 3A via the 10k resistor a pulse of 30000 volts is required.

If the 10k resistor had a high enough rating a simple 10k resistor and 4N25 input could sustain
600 volts at the input indefinitely (60mA). As the input is for very low speed switching a capacitor on the input may be worthwhile.
More protection devices could be employed as many have suggested.

Finding a suitable resistor may be the hardest part it needs a high voltage rating and a high pulse rating.

[ken_k]

Hi some more information.

https://www.smd.ru/upload/medialibrary/a3d/iso_7637_2.pdf
pages 16 and 17 interesting.

A pulse of 350 millisecond of 200V gives a peak wattage of 4 watts into 10K with a duration of 0.35 seconds that is 1.4 joules. Note the area under the curve will be significantly lower power.
I think this is quite doable.

I don't think this is any more than a rough guide. I would not use <1W resistor.
https://calculator.academy/resistor-pulse-power-calculator/

One could set up all the waveforms listed in ISO7637-2 and test them in a spice package.

Little fuse has spice models for many of its TVS diodes.

https://www.littelfuse.com/technical-resources_old/spice-models/tvs-diode-spice-models.aspx

Always fun to blow up virtual components.

[ken_k]

Sorry all - my previous post was about protecting the entire supply to this piece of electronics rather than just the logic inputs.  Apologies for any confusion caused.

David

Hi David you have brought up a very good point, protecting the 12/24V supply into any mobile electronic device according to ISO 7632-2 is a far more demanding task. The high voltage FET switch may be an option. I'm guessing a lot of consumer product would fail the load dump test.

some data
https://www.analog.com/en/technical-articles/loaddump-protection-for-24v-automotive-applications.html

[david]

Ken - as you rightly pointed out, high impedance inputs are easier to protect than main 24V power supply inputs which may need to draw ampere level currents.
Equipment supplied at OEM level may need to meet the individual vehicle manufacturers standard which is usually more severe than the ISO standards pertaining to the device.  Aftermarket supplied equipment is more of a lucky dip but even trivial items such as cigarette lighter power adaptors are covered by UL 2089 if sold in the US. 
Load dump testing when applied to power inputs can be challenging and we found the combo of a short term transient protector and a FET series switch proved to be very robust. It will depend very much on the supply current used as any series resistance that can be allowed makes the job easy and provides something for capacitors to filter with for the high voltage spikes. 

Years back we supplied marine GPS equipment which proved very reliable except for a small region in the Philippines. Confused why the problem should be so isolated they eventually sent a rep to visit and he found that the boss of a small fishing fleet would drive to the wharf and the battery was taken out of his car and used to start the first boat.  Once the motor was going the battery was then moved to the second boat and so on.

Should we mention ESD testing?  Very short duration but very high voltage and very fast risetimes.  Applied to all input and output wiring and also any metal on the housing - especially screws on a plastic case that may pass through a pcb. 

Cheers,
David

[ken_k]

Should we mention ESD testing?  Very short duration but very high voltage and very fast risetimes.  Applied to all input and output wiring and also any metal on the housing - especially screws on a plastic case that may pass through a pcb.

https://www.testups.com/iso-10605-esd-test-setup-requirements/

https://www.nutwooduk.co.uk/archive/keitharmstrong/design_techniques6.html

Even a simple capacitor resistor network can be very useful as a pre-compliance test. If it passed the HV spark test it will probably be OK.

David a good story re the boat battery.

[John Lawton]

Ken,
many thanks, especially for the Keith Armstrong piece which is excellent stuff.

As you say a simple resistor + capacitor circuit can be adequate protection against ESD for PIC inputs which surprised me when I found that out.

[david]

Great link Ken.

I guess after passing ESD tests with flying colours it's on to the Susceptibility and Emissions Testing.  Nothing like a few nervous hours as your product is put through TEM Cell tests.

Cheers,
David

[John Lawton]

The nice thing with (PIC) microcontrollers that don't have the old fashioned multibit address and data buses of the older micros is the much lower levels of EMI they produce. I've never had a problem (i.e. have undetectable emissions) with these tests.

John